A Local/Regional Based CAD System for Early Diagnosis of Alzheimer's Disease Using sMRI Scans

Abstract

Alzheimer's disease (AD) is one of the neurodegenerative diseases, an irreparable one, that targets the central nervous system and causes dementia. During its progression, the disease goes through a number of stages where the diagnosis of the disease at its early stage is highly recommended. Despite this recommendation, accomplishing this diagnosis task faces a number of obstacles including the variable impact of the disease on its sufferers. This paper mainly aims to assist in the early diagnosis process of AD through introducing a brain regional based computer-aided diagnosis (CAD) system, using structural magnetic resonance imaging (sMRI), that goes through four main stages: preprocessing, brain labeling, extracting the discriminant features, as well as diagnosing. The novelty of the our work is to offer a local/regional diagnosis to serve the subject-dependent effect of the disease followed by a global diagnosis with an overall promising performance as evaluated with the related work. The experimental results show accuracy of 96.6%, specificity of 100%, and sensitivity of 94.25%. Validating our system with the related work and some well-known classifiers shows promising results in addressing this research point.